The ultimate Fabric, Power BI, SQL, and AI community-led learning event. Save €200 with code FABCOMM.
Get registeredCompete to become Power BI Data Viz World Champion! First round ends August 18th. Get started.
I have a strange issue at hand. I have a report that's built off a fact table that's supported by direct query connected to my sql warehouse in databricks. Additionally, I have a number of different date tables that are feeding in data from that same warehouse and connected to my fact table through an import via dataflow.
Issue I'm running into is when I'm filtering data of transactions say for a single day, the report gets thrown an error saying I've hit a 1M row error, which should be impossible because running this same query manually in the db reveals I should receive no more than 500 rows. I took a look at the query that was pinging the db from the reports, and the report filter for date wasn't being included, although the other filter parameters are being included.
Hey @vegasde100,
This is a classic DirectQuery + Import mixed model issue. Here's what's likely happening and how to fix it:
Root Cause: Your date filter isn't being pushed down to the DirectQuery fact table because Power BI can't bridge filters between Import (date tables) and DirectQuery (fact table) modes in the same model.
Quick Fixes:
The 1M row limit suggests the query is pulling the entire fact table without the date filter being applied. Start with #2 (switching date tables to DirectQuery) - that's usually the fastest solution for this scenario.
Worth checking your relationship cardinality settings too - sometimes incorrect many-to-many relationships can cause filter propagation issues.
Fixed? ✓ Mark it • Share it • Help others!
Best Regards,
Jainesh Poojara | Power BI Developer
Appreciate the explanation, Jainesh. I'll look into changing the date tables from Import to DirectQuery mode.
One thing that still puzzles me is that, initially what I attempted to do was switch the date slicer field reference from the date table to the date field in the fact table, which still seems to have thrown the 1M row error; any idea on that one?